Detecting and Measuring Depression on Social Media Using a Machine Learning Approach: Systematic Review
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
BACKGROUND: Detection of depression gained prominence soon after this troublesome disease emerged as a serious public health concern worldwide. OBJECTIVE: This systematic review aims to summarize the findings of previous studies concerning applying machine learning (ML) methods to text data from social media to detect depressive symptoms and to suggest directions for future research in this area. METHODS: A bibliographic search was conducted for the period of January 1990 to December 2020 in Google Scholar, PubMed, Medline, ERIC, PsycINFO, and BioMed. Two reviewers retrieved and independently assessed the 418 studies consisting of 322 articles identified through database searching and 96 articles identified through other sources; 17 of the studies met the criteria for inclusion. RESULTS: Of the 17 studies, 10 had identified depression based on researcher-inferred mental status, 5 had identified it based on users' own descriptions of their mental status, and 2 were identified based on community membership. The ML approaches of 13 of the 17 studies were supervised learning approaches, while 3 used unsupervised learning approaches; the remaining 1 study did not describe its ML approach. Challenges in areas such as sampling, optimization of approaches to prediction and their features, generalizability, privacy, and other ethical issues call for further research. CONCLUSIONS: ML approaches applied to text data from users on social media can work effectively in depression detection and could serve as complementary tools in public mental health practice.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.005 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it